From 7fa1c73b010989a19ee7b6bce939a2f51d9f2a8b Mon Sep 17 00:00:00 2001
From: Remi PLANEL <rplanel@pasteur.fr>
Date: Fri, 26 Apr 2024 17:53:35 +0200
Subject: [PATCH] operon strcture: take gene size into account
---
components/OperonStructure.vue | 98 ++++++++++++++-----
.../df-wiki-cli/df_wiki_cli/content/main.py | 23 ++++-
.../df_wiki_cli/meilisearch/update/main.py | 1 +
packages/df-wiki-cli/pyproject.toml | 2 +-
types/structure.ts | 2 +
5 files changed, 98 insertions(+), 28 deletions(-)
diff --git a/components/OperonStructure.vue b/components/OperonStructure.vue
index 2fa2077f..da343553 100644
--- a/components/OperonStructure.vue
+++ b/components/OperonStructure.vue
@@ -14,7 +14,7 @@ const structureBasket = useStructuresBasket()
const props = withDefaults(defineProps<Props>(), {
genes: null
});
-const height = ref<number>(120)
+const height = ref<number>(200)
const svgRef = ref<SVGElement | null>(null)
const margin = ref<PlotMargin>({
marginTop: 10,
@@ -35,16 +35,36 @@ const domain = computed(() => {
return genes?.map(d => { return d.gene })
})
+const innerPadding = ref<number>(0.5)
+
+const totalGeneLength = computed(() => {
+ const genes = toValue(computedGenes)
+ return genes.reduce((acc, curr) => {
+ return acc + (curr?.size ?? 10) + innerPadding.value
+ }, 0)
+})
+
+
+const domainGenes = computed(() => {
+ return [0, totalGeneLength.value]
+
+})
const xScale = computed(() => {
return d3.scaleBand()
- .paddingInner(0.1)
+ .paddingInner(0)
.domain(toValue(domain))
.range([0, computedPlotWidth.value])
})
+const xScaleGenes = computed(() => {
+ return d3.scaleLinear()
+ .domain(toValue(domainGenes))
+ .range([0, computedPlotWidth.value])
+})
+
const yScale = ref(d3.scaleBand()
- .domain(['img', 'gene'])
+ .domain(['img', 'buff', 'buff2', 'gene'])
.range([toValue(margin).marginTop, toValue(height)]));
const gbContainer = ref(null)
const computedContainerWidth = computed(() => {
@@ -60,9 +80,15 @@ const computedPlotWidth = computed(() => {
const computedGenes = computed<StructureOperonGene[]>(() => {
const genes = toValue(props.genes)
if (genes !== null) {
+ let currentSumSize = 0
return genes.map(d => {
+ const size = d?.size ?? 10
+ const position = currentSumSize
+ currentSumSize = position + size + innerPadding.value
return {
...d,
+ size,
+ position
}
})
}
@@ -83,14 +109,14 @@ const structureVersion = computed(() => {
const genesWithCoord = computed<StructureOperonGeneWithCoordinate[]>(() => {
const genes = toValue(computedGenes)
- const xScaleVal = toValue(xScale)
+ const xScaleVal = toValue(xScaleGenes)
const yScaleVal = toValue(yScale)
if (genes !== null) {
return genes.map(d => {
return {
...d,
- width: xScaleVal.bandwidth(),
- x: xScaleVal(d.gene),
+ width: xScaleVal(d.size),
+ x: xScaleVal(d.position),
y: yScaleVal('gene'),
height: yScaleVal.bandwidth()
}
@@ -127,8 +153,6 @@ function draw() {
gx.call(g => g.select(".domain")
.remove())
-
-
.selectAll("text")
.attr("transform", 'rotate(20)')
.attr("text-anchor", "start")
@@ -143,19 +167,24 @@ function draw() {
function drawGenes(genesGroup: d3.Selection<SVGElement, any, SVGElement, any>) {
const data = toValue(genesWithCoord)
const genesSelection = genesGroup
- .selectAll("g.gene") // get all "existing" lines in svg
+ .selectAll("g.operon-item") // get all "existing" lines in svg
.data<StructureOperonGeneWithCoordinate>(data) // sync them with our data
.join(
enter => {
-
-
- const gGene = enter.append("g")
- .classed("gene", true);
-
- gGene.append("path")
+ const gOperonItem = enter.append("g")
+ .classed("operon-item", true);
+
+ // gene grp
+ const gGene = gOperonItem.append("g")
+ .classed("gene-grp", true)
+ gGene
+ .append("path")
.classed("gene", true)
- gGene.append("g").classed("img", true).append("image")
+ // img group
+ gOperonItem
+ .append("g").classed("img", true)
+ .append("image")
.on("mouseover", function (event) {
const target = d3.select(event.srcElement.parentElement)
target
@@ -170,32 +199,51 @@ function drawGenes(genesGroup: d3.Selection<SVGElement, any, SVGElement, any>) {
// .attr("stroke", null)
.attr("cursor", "unset")
})
- gGene.append("text")
+
+
+ gOperonItem.append("text")
// .attr("fill", "white")
.classed("gene-label", true)
.attr("fill", "currentColor")
.attr("dominant-baseline", "middle")
- gGene.append("title")
- return gGene
+
+
+ gOperonItem.append("line")
+ gOperonItem.append("title")
+ return gOperonItem
},
update => update,
exit => exit.remove()
)
- genesSelection.attr("transform", d => `translate(${d.x}, 0)`)
- genesSelection.select("g.img").select("image")
+ genesSelection.select("g.gene-grp").attr("transform", d => `translate(${d.x}, 0)`)
+ genesSelection.select("g.img")
+ .attr("transform", d => `translate(${xScale.value(d.gene)})`)
+ .select("image")
.attr("transform", d => `translate(0, ${toValue(yScale)("img")})`)
.attr("href", d => d?.structImg ?? null)
- .attr("width", d => d.width)
- .attr("height", d => d.height)
+ .attr("width", toValue(xScale).bandwidth())
+ .attr("height", toValue(yScale).step() * 3)
+ .attr("preserveAspectRatio", "xMidYMid meet")
.on("click", function (event) {
const data = d3.select<SVGElement, StructureOperonGeneWithCoordinate>(this).data()
structureBasket.set(data.map(s => s?.structPath ?? ''))
})
- genesSelection.select("path.gene")
+
+
+ genesSelection.select("g.gene-grp").select("path.gene")
.attr("transform", d => `translate(0, ${d.y})`)
- .attr("fill", d => color(d.gene))
+ .attr("fill", d => color(d.system))
.attr("d", d => drawGene(d).toString())
+
+ genesSelection.select("line")
+ // x1="0" y1="80" x2="100" y2="20" stroke="black"
+ .attr("x1", d => xScale.value(d.gene) + xScale.value.bandwidth() / 2)
+ .attr("y1", d => yScale.value("buff2") + yScale.value.bandwidth() / 3)
+ .attr("x2", d => xScaleGenes.value(d.position) + xScaleGenes.value(d.size / 2))
+ .attr("y2", d => yScale.value("gene") - 2)
+ .attr("stroke", "currentColor")
+
}
function drawGene({ width, height }) {
diff --git a/packages/df-wiki-cli/df_wiki_cli/content/main.py b/packages/df-wiki-cli/df_wiki_cli/content/main.py
index c3373ef5..fda38afe 100644
--- a/packages/df-wiki-cli/df_wiki_cli/content/main.py
+++ b/packages/df-wiki-cli/df_wiki_cli/content/main.py
@@ -316,6 +316,18 @@ def system_operon_structure(
"version": def_path["version"],
"exchangeables": None,
}
+
+ # get hmm definition to extract the size
+ base_path_list = def_path["path"].parts[0:-4]
+ profile_path = (
+ Path(*base_path_list)
+ / f"profiles/{current_gene['gene']}.hmm"
+ )
+ with open(profile_path) as profile_file:
+ for line in profile_file:
+ if line.startswith("LENG"):
+ size = int(re.split(r"\s+", line)[1])
+ current_gene["size"] = size
system_genes.append(current_gene)
if elem.tag == "gene" and in_exchangeables:
exchangeables.append(elem.attrib["name"])
@@ -329,7 +341,15 @@ def system_operon_structure(
exchangeables = []
with open(output, "w") as f:
- fieldnames = ["id", "system", "subsystem", "version", "gene", "exchangeables"]
+ fieldnames = [
+ "id",
+ "system",
+ "subsystem",
+ "version",
+ "gene",
+ "size",
+ "exchangeables",
+ ]
writer = csv.DictWriter(f, fieldnames=fieldnames)
writer.writeheader()
for id, gene in enumerate(system_genes):
@@ -727,7 +747,6 @@ def find_model_definition(system, subsystem, list_paths):
parts = path.parts
if path.stem == subsystem and parts[-2] == system:
console.rule(f"{system} - {subsystem}")
- console.print(p)
found_path = {"path": path, "version": p["version"]}
break
diff --git a/packages/df-wiki-cli/df_wiki_cli/meilisearch/update/main.py b/packages/df-wiki-cli/df_wiki_cli/meilisearch/update/main.py
index de579e18..a219b235 100644
--- a/packages/df-wiki-cli/df_wiki_cli/meilisearch/update/main.py
+++ b/packages/df-wiki-cli/df_wiki_cli/meilisearch/update/main.py
@@ -110,6 +110,7 @@ class SystemOperonStructure(BaseModel):
subsystem: str
version: str
gene: str
+ size: int
exchangeables: Optional[List[str]]
diff --git a/packages/df-wiki-cli/pyproject.toml b/packages/df-wiki-cli/pyproject.toml
index db1e9ac2..fd210805 100644
--- a/packages/df-wiki-cli/pyproject.toml
+++ b/packages/df-wiki-cli/pyproject.toml
@@ -1,6 +1,6 @@
[tool.poetry]
name = "df-wiki-cli"
-version = "0.2.1"
+version = "0.2.2"
description = ""
authors = ["Remi PLANEL <rplanel@pasteur.fr>"]
readme = "README.md"
diff --git a/types/structure.ts b/types/structure.ts
index e7d08dcb..e622d92c 100644
--- a/types/structure.ts
+++ b/types/structure.ts
@@ -5,6 +5,8 @@ export interface StructureOperonGene {
version: string
system: string
exchangeables: string[]
+ size: number
+ position: number
}
--
GitLab